Numerous studies have sought to elucidate the mechanisms driving the assembly and distribution of biodiversity, from large-scale climatic gradients to local-scale constraints on biotic competition, especially the role of forest structure in influencing micro-site conditions and mediating resource competition. However, while empirical findings on the relationship between biodiversity and forest structure from local studies are promising, conclusions have limited utility for inference into larger-scale patterns across vastly different environmental and climatic settings. To address this knowledge gap, this study harnesses the extraordinary scope of the entire NEON terrestrial dataset to answer the following questions: what are the independent and interactive effects of forest structure and climate in constraining multi-taxon diversity across continuous bioclimatic gradients of North America? What are the implications of the shape of this interaction for understanding the primary mechanism of community assembly? Statistical methods rely on ensemble machine learning for insight into predictor importance and model accuracy, as well as Bayesian hierarchical Generalized Linear Mixed Models (GLMMs) for inference into mechanism. Specifically, we test the hypothesis that three global-scale bioclimatic drivers of diversity - climatic energy, stress, and stability – interact significantly with local-scale field and remotely-sensed forest structural attributes to drive the assembly of multi-taxa diversity.
Results/Conclusions
Model results confirm the complementarity of structure and climate in predicting plant, tree, bird, and small mammal diversity, with cross-validated adj-R2s of 0.70, 0.68, 0.87, and 0.63, respectively. Compared with models predicting diversity from one forest structural attribute, the addition of single climate predictors increased model accuracy 0.1-0.4 (adj-R2). GLMM model results partly confirm our hypothesis that increased temperatures (e.g. MAT, PET, and GDD), decreased cold stress, and greater climatic stability (e.g. variation in GPP) accentuate the positive relationship between structure and plant/tree richness. Precipitation patterns, on the other hand, did not fully conform to expectations, with drought and heat stress positively interacting with forest structure, and precipitation (annual totals and seasonality) exhibiting significant negative interactions. Taken together, these results advance our understanding of the interaction of large- and small-scale constraints on the multi-scale, multi-taxa assembly diversity across North America. In addition, they provide an ecological basis for using space-borne LiDAR and hyperspectral sensors to predict biodiversity patterns at regional to global scales.